Will AI Let Teams Scale?

One engineer, given enough time and resources, can solve almost anything. The trouble starts the moment you need a second one. After that, you are not solving a problem anymore. You are building a team that can keep solving it.

On your own, the constraints are simple: your time, your focus, your willingness to learn a new language or go as deep into a domain as the work demands. Add people and the whole shape of the job changes. Now you need a structure, an agreed way of working, a way to deliver quality that repeats rather than depending on heroics, and enough room inside all of it for people to actually grow. None of that is the problem you were originally hired to solve. All of it is now your problem.

The Skill Set Exploded

Set aside, for a moment, the AI skills everyone is now scrambling to acquire. Even without them, the job has grown almost past recognition. An engineer is no longer the person who writes the code. They are expected to understand how software is designed, developed, built, tested, deployed, monitored, and then evolved safely over years, frequently all at the same time. In SaaS especially, where the application is always on, always changing, serving many tenants at once, and expected to grow without ever going down, you are not really shipping features anymore. You are stewarding a living system while it runs.

Here is the tension I keep returning to. The technology has scaled. The tooling, the platforms, the abstractions, the sheer reach of what a small group can now ship, all of it has scaled enormously. The people building it have not scaled the same way, because people do not scale like infrastructure. You cannot autoscale judgment, or trust, or the slow accumulation of shared context that makes a team good. When the system grows faster than the people responsible for it, you are quietly breeding a catastrophe, and it tends to stay quiet right up until it is not.

Teams Are the Bottleneck You Cannot Provision Away

Which is why I have come to believe the hardest part of software engineering today has almost nothing to do with software. It is aligning people, balancing speed against quality without lying to yourself about either, growing skills while still shipping reliably, designing systems that survive the people who built them moving on, and keeping a culture where the learning never actually stops. I have made the case before for why strong teams fix weak processes and laid out the philosophy I try to build every team on, so I will not repeat it here. The short version is that building teams which deliver sustainably is the real craft, and it runs on leadership, empathy, structure, and patience far more than on any framework.

Does AI Close the Gap, or Widen It?

What I genuinely do not know yet is how all of this collides with AI. The optimistic version is seductive: AI absorbs the toil, lifts the floor on what one person can do, and finally lets a team scale closer to the rate the technology already moves at. The skill explosion gets compressed back down into something a human-sized team can actually hold. The pessimistic version is just as plausible. AI raises the ceiling on what a team is expected to ship without raising the ceiling on trust, alignment, or the patience it takes to grow a junior into a senior. The output scales, the human bottleneck does not move, and the gap I described only gets wider, faster. We learned to code so that others could type a prompt; what that does to the team underneath the prompt is still being written.

So these are the two questions I am sitting with, and I do not have clean answers to either. The first is how team management itself changes when a large share of the work is done with, or by, AI. The second is harder: whether AI can help us deliver the quality and the repeatable outcomes we need over time while still leaving people the room to grow, or whether optimising hard for the first quietly eats the second. A team that ships flawlessly and learns nothing is not a team that lasts.